The main hypothesis is that complex cooperative behavior arises through a principled interaction between emotion and communication. A primary emotion such as fearfulness results in avoiding the risky actions necessary for the task. This emotion is overcome through a second emotion, affiliation to the team. This emotion also enhances communication, which would otherwise be obscured by fear, so that the actions can be effectively coordinated. To evaluate the hypothesis, a computer scientist will join forces with a behavioral biologist in the proposed project. The approach is based on the principle that, to understand such behavior, it is useful to understand its evolutionary origins. This perspective leads to a unique methodology: using computational evolution to analyze and draw conclusions from an animal model. Specifically, the project focuses on the complex cooperation that emerges in spotted hyenas of Eastern Africa (Crocuta crocuta) when they compete for food resources with larger and more powerful animals, i.e. lions. The specific aims are to: (1) Characterize the target behavior and its endocrine correlates in detail in the animal model; (2) build a computational model that replicates the behavior; (3) determine how the behavior is produced in the model; (4) understand the evolutionary origins of the behavior; and (5) determine how problems with the behavior can be ameliorated. The end result will be a computationally verified theory of how emotions and communication mediate sophisticated cooperative behavior in mammals, and why it occurs this way. This theory can ultimately be used to suggest therapeutic measures for accommodating humans who have trouble in cooperative tasks, including those who are excessively withdrawn or dominant, those with at or excessive emotions, and those with communication disorders, as well as strategies for controlling mob violence. Such insights are difficult to obtain in general, but the unique combination of an animal model and evolutionary computation in the proposed project should make it possible.